The Research and Application of Content-Based Multi-feature Comprehensively Weighting Video-Image Retrieval Algorithm

The problems of accuracy and computational complexity in extracting image features(color, texture and shape) in traditional Image retrieval algorithm result in a bigger error of image retrieval result and the lower efficiency of retrieval. A Content-Based Multi-Feature Comprehensively Weighting Video-Image Retrieval Algorithm is proposed to settle the problem. The essence of the algorithm is, set an appropriate threshold value and calculate the similarity of the color feature between the target image and each retrieval image to determine whether the calculation of shape feature is necessary, and similarly, determine whether the calculation of texture feature is necessary. Finally, compare the similarity distance of chosen image to get the final result. The algorithm is experimented for many times, comparing the extraction method of the three image features during the tests, and meanwhile designed a set of image retrieval system basing on the C#+EMGUCV platform, in the end, verifying the efficiency and accuracy of the image retrieval algorithm that this article have proposed.